Upload
others
View
14
Download
0
Embed Size (px)
Citation preview
Albanian Word Stress
Manchester Phonology Meeting, May 20-22, 2004
Jochen Trommer Angela GrimmInstitute of Cognitive Science Department of General Linguistics
University of Osnabruck University of Groningen
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Overview
• Basic features of Albanian word stress
• Previous approaches
• Empirical evaluation
• An OT-analysis
1
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Basic Features of Albanian Word Stress
• 1 stress/word form
• Paradigmatic uniformity
• Right-edge orientation
• Sensitivity to vowel quality
• Sensitivity to syllable weight
3
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
1 Stress/Word Form
from the left from the right1st ma.l-i ‘the mountain’ vesh.ti.re.si ‘difficulty’2st nje.ri ‘human’ per.pu.ni.m-i ‘the treatment’3rd per.pa.rim ‘progress’ ge.njesh.tr-a.ve ‘of lies’4th qy.te.te.rim ‘civilization’ kum.bu.ll-a.ve ‘of the plums’
4
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Paradigmatic Uniformity
babo, ’midwife’, agon ’it dawns’
(C)VCV ba.bo Nom. Sg. indef. a.go Imperative(C)VCVC ba.bon Acc. Sg. def. a.gon Pres. 3sg(C)VCVCCV ba.bo.ja Nom. Sg. def. a.go.je Impf. 2sg
➔ Inflection doesn’t affect stress placement
➔ Extends to adverbial suffixes and prefixes
5
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Right-Edge Orientation
Inflected Forms
from the left from the right1st ma.l-i ‘the mountain’ vesh.ti.re.si ‘difficulty’2st nje.ri ‘human’ per.pu.ni.m-i ‘the treatment’3rd per.pa.rim ‘progress’ ge.njesh.tr-a.ve ‘of lies’4th qy.te.te.rim ‘civilization’ kum.bu.ll-a.ve ‘of the plums’
Base Forms
1st mal ‘mountain’ vesh.ti.re.si per.pu.nim2st nje.ri ‘human’ ge.njesh.ter kum.bull3rd per.pa.rim ‘progress’4th qy.te.te.rim ‘civilization’
6
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Sensitivity to Vowel Quality
Final Vowel = Stress =e a.ne, he.ne, e.he, pro.ne, si.vee fa.qe, en.dje, en.de, go.lle, fi.ce penultimao ba.bo, ne.to, lo.co, bir.koa ha.ta, pas.ter.ma, xhe.la, ot.ra, ri.xhai ba.ri, gju.he.si, qer.shi, kom.shi, zi.li finalu ash.tu, a.ke.ku ku.cu.ru
➔ Only quality of vowels in final syllables is “visible“
➔ High and low vowels attract stress
7
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Sensitivity to Syllable Weight
Final Syllable
open closed Stressqer.shi ‘cherry’ ar.mik ‘enemy’ha.ta ‘calamity’ re.zul.tat ‘result’ finalash.tu ‘this way’ ci.fut ‘gipsy’fa.qe ‘face ’ she.qer ‘sugar’ finalba.bo ‘midwife’ pa.tok ‘gander’a.ne ‘side’ a.fer ‘near’ penultimate
➔ Only Weight of final syllables is “visible“
➔ Closed Syllables with full vowels attract stress
8
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Stress Algorithm
For a given word form W :
1. Find the base form B of W ’s inflectional paradigm
2. If Monosyllabic(B): ➔ stress only syllable of B
3. Else: Find the final syllable S of B
a. If Full Vowel(Nucleus(S)) and Closed(S)or Nucleus(S) = i,a,u:
➔ stress final syllable of B
b. Else: ➔ stress penultimate syllable of B
9
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Three Example Applications
Input: gju.he.si ge.njesh.tra.ve pu.no.ni‘linguistics’ ‘of lies (abl.pl)’ ‘you (pl.) work’
1. Base Form gju.he.si ge.njesh.ter pu.non2. Monosyllabic3. Final Syllable si ter nona. Closed + Full Vowel non
or i,a,u ➔ Final sib. Else ➔ Penultimate njeshOutput: gju.he.si ge.njesh.tra.ve pu.no.ni
10
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Dodi and Gjinari (1983)
“the position of stress in Albanian wordscannot be determined by a general rule”. (p. 129)
from the left from the right1st ma.l-i ‘the mountain’ vesh.ti.re.si ‘difficulty’2st nje.ri ‘human’ per.pu.ni.m-i ‘the treatment’3rd per.pa.rim ‘progress’ ge.njesh.tr-a.ve ‘of lies’4th qy.te.te.rim ‘civilization’ kum.bu.ll-a.ve ‘of the plums’
But this neglects:
• The difference between derivation and inflection
• The predictability of most stress assignment
12
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Newmark et al. (1982)
“in general, the main stress in an Albanian stemfalls on its last syllable,the main stress of an Albanian word . . .
falls on its last stem . . . (p. 15).
This doesn’t capture:
• Penultimate stress
• Sensitivity to syllable weight and vowel quality
13
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Bevington (1974)
(1) V ➔ [+stress] C0
⟨
eao
e(C)asull
⟩
] Stem<-V>
Problems:
• Rule doesn’t refer to natural class
• Reference to syntactic categories (“-V”)
14
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Maynard (1997)
(1) Input: pu.ne.-tor, ‘worker’
1-STRESS FAITH-STRESS STRESSWORD I/O RIGHT-EDGE
pu.ne-.tor *!pu.ne-.tor * *!
☞ pu.ne-.tor *pu.ne-.tor **!
Doesn’t capture:
• Penultimate stress
• Sensitivity to syllable weight and vowel quality
15
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Number of Stems with n Syllables in Annotated Data
n=1 n=2 n=3 n=4 n=5 n=6 n=7 n=8 n=9 SumVerbs 0 1496 1231 273 66 11 0 1 0 3078Nouns 0 3762 3164 1507 348 66 10 5 0 8862Adj. 0 1794 1666 723 162 40 11 3 0 4399Adv. 0 374 226 103 22 0 0 0 0 725Prep. 0 12 0 1 0 0 0 0 0 13All 0 7438 6287 2607 598 117 21 9 0 17077
17
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Distribution of Stress Positions
syllablenumber Stress Positionof word 1 2 3 4 5 6 7 82 2820 46053 33 1789 44754 3 9 1167 14305 6 195 3986 43 747 5 168 1 8
18
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Vowel Quality in Vowel-Final Stems and Final Stress
a i u o efinal 65 (78.3%) 991 (93.6%) 10 (66.7%) 25 (29.1%) 104 (7.9%)non-final 18 (21.7%) 68 (6.4%) 5 (33.3%) 56 (70.9%) 1206 (92.1%)all 83 1059 15 86 1310
19
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Coda Consonants and Final Stress
Final Syllables with Full Vowel
No Coda-C 1 Coda-C 2 Coda-Cfinal 1199 (46.9%) 8663 (93.8%) 1084 (99.6%)non-final 1358 (53.1%) 775 (6.2%) 4 (0.4%)all 2557 9238 1088
Final Syllables with Schwa
No Coda-C 1 Coda-C 2 Coda-Cfinal 42 (1.7%) 49 (15.5%)non-final 2474 (98.3%) 9 (84.5%)all 2517 58
20
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Evaluation of Different Theories
Proposed Algorithm
right 16247 (97.7%)wrong 378 (2.3%)
Simple Theories on Albanian Stress
initial second penultimate finalright 2856 (16.7%) 6403 (37.5%) 6020 (35.3%) 11006 (64.4%)wrong 14221 (83.3%) 10674 (62.5%) 11057 (64.7%) 6071 (35.6%)
21
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Right-Edge Orientation and Restriction to 1 Stress
(1) Final Bi-syllabic Trochee
Align(Foot,R,PWd,R)
Trochee
☞ a. di.te.(lind.je)☞ b. di.te.lind.(je)
c. di.(te.lind).je *!d. (di.te.) lind.je *!*e. (di.te.)(lind.je) *!*
(2) Final Mono-syllabic Trochee
Align(Foot,R,PWd,R)
Trochee
☞ a. ar.(gjend)☞ b. (ar.gjend)
c. (ar).gjend *!
23
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Sensibility of Stress to Vowel Quality
a, i, u > e, o > ehigh and low non-centralized centralizedvowels mid vowels mid voweltend to be tend not to be tends not to bestressed in all stressed in open stressed in allfinal syllables final syllables final syllables
24
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Formalizing the Notion “Unstressed Vowel”:(Non-)Designated Terminal Elements (de Lacy, 2002)
4Ft: A segment dominated through an uninterrupted lineof strong nodes by a foot node
∗Non-4Ft: A segment dominated by a foot node and not a 4Ft
Word
|
σ
Tw es
Fs
σs
rw is
σw
nw es
Non Non Non4Ft 4Ft 4Ft 4Ft
25
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Sensibility to Vowel Quality by Constraints on Non-DTEs
a,i,u > e,o > e
∗Non-4Ft/{a,i,u} � ∗Non-4Ft/{e,o} � ∗Non-4Ft/{e}
a,i,u should not e,o should not e should notbe Non-DTE’s be Non-DTE’s be a Non-DTE
If a,i,u are part of a foot If e,o are part of a foot If e is part of a footthey should be stressed they should be stressed it should be stressed
26
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Sensibility to Vowel Quality
(1) bari, ‘lawn’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o} {e}☞ a. ba.(ri) *
b. (ba.ri) *!
(2) babo, ‘midwife’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o} {e}a. ba.(bo) *!
☞ b. (ba.bo) *
(3) bale, ‘ball’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o} {e}a. ba.(le) *!
☞ b. (ba.le) *
27
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Weight by Position (Rosenthal and van der Hulst, 1999)
(1) qershi, ‘cherry’
Append *µ/Cons☞ a. qerµ.shi *
b. qµerµ.shi **!c. qµer.shi *! **d. qer.shi *!
(2) qershi, ‘cherry’
*µ/Cons Appenda. qerµ.shi *!b. qµerµ.shi *!*c. qµer.shi *!* *
☞ d. qer.shi *
28
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Weight by Position by Position (Rosenthal and van der Hulst, 1999)
(1) sheqer, ‘sugar’
Ft-Bin ∗Non-4Ft/ {e,o} *µ/Cons Appenda. (she.qer) *! *b. she.(qer) *! *
☞ c. she.(qerµ) *
(2) serbe, ‘Serbian’
Ft-Bin ∗Non-4Ft/ {e,o} *µ/Cons Append☞ a. (ser.be) * *
b. (serµ.be) * *!
29
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Sensibility to Syllable Weight
(1) babo, midwive
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}a. ba.(bo) *!
☞ b. (ba.bo) *
(2) patok, ‘gander’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}☞ a. pa.(tokµ) *
b. pa.(tok) *! *c. (pa.tok) *!
30
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Putting it All Together (Vowel-Final Forms)
(1) bari, ‘lawn’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}☞ a. ba.(ri) *
b. (ba.ri) *!
(2) babo, midwive
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}a. ba.(bo) *!
☞ b. (ba.bo) *
(3) bale, ‘ball’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}a. ba.(le) *!
☞ b. (ba.le) *
31
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Putting it All Together (Consonant-Final Forms)(1) besnik, ‘true’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}☞ a. bes.(nikµ) *
b. (bes.nik) *! *
(2) patok, ‘gander’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}☞ a. pa.(tokµ) *
b. (pa.tok) *! *
(3) afer, ‘near’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}a. a.(ferµ) *!
☞ b. (a.fer) * *
32
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Summary
• Albanian word stress is sensitive to vowel quality and syllable weight
• Both properties are only visible in last syllable of inflectional base forms
• Captures more than 97% of Albanian vocabulary
• Principled implementation by prosodic hierarchy and violable constraints
33
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Theoretical Implications
• Evidence for constraint-based analysis of weight-by-position by position
• So far undocumented vowel prominence system in stress assignment
34
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Markedness Hierarchies for Stressed Vowels
Albanian: a i,u > e,o > e
Kenstowicz (1996): a > e,o > i,u > e
de Lacy (2002): a > e,o > i,u > e > i,u
➔ Albanian vowel prominence corresponds not to sonority,but to peripherality or distinctness
➔ Link to preferences in minimal vowel systems
➔ Specific phonetic properties of high vowels in Albanian?
35
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Paradigm Uniformity by Output-Output Constraints
(1) Base Form: argjend, ‘silver’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}☞ a. ar.(gjendµ) *
b. (ar.gjend) *! *
(2) Derived Form: argjendi, ‘the silver’
OO-Ident-Stress Phonology☞ a. ar.gjen.di *
b. ar.gjen.di *! *c. ar.gjen.di *!
OO-Ident-Stress: If segment Sd in a derived form corresponds to Sb in a base formSd is the DTE of syllable σd, and Sb the DTE of σb, thenσd is stressed if σb is.
36
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Capturing Paradigm Uniformity by Cyclic Input-Output Constraints
(1) Stem: argjend, ‘silver’
AlignRt∗Non-4Ft/
∗Non-4Ft/∗Non-4Ft/
Trochee {a,i,u}Ft-Bin
{e,o}*µ/Cons Append
{e}☞ a. ar.(gjendµ) *
b. (ar.gjend) *! *
(2) Derived Form: argjendi, ‘the silver’
IO-Ident-Stress Phonology☞ a. ar.gjen.di *
b. ar.gjen.di *! *c. ar.gjen.di *!
IO-Ident-Stress: If segment Sd in a derived form corresponds to Sb in a base formSd is the DTE of syllable σd, and Sb the DTE of σb, thenσd is stressed if σb is.
37
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Are there vowel-final Verb Stems?
final C final C final V final V ?1sg hap-Ø vras-Ø pi-Ø formo-j2sg hap-Ø vret-Ø pi-Ø formo-n3sg hap-Ø vret-Ø pi-Ø formo-n1pl hap-im vras-im pi-me formo-jme2pl hap-ni vris-ni pi-ni formo-ni3pl hap-in vras-in pi-ne formo-jne
➔2/3sg is a plausible base form
38
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
Alternative Analysis of Vowel-Final Stems
final C final C final V final V ?1sg hap-Ø vras-Ø pi-Ø formoj-Ø2sg hap-Ø vret-Ø pi-Ø formon-Ø3sg hap-Ø vret-Ø pi-Ø formon-Ø1pl hap-im vras-im pi-me formoj-me2pl hap-ni vris-ni pi-ni formon-ni3pl hap-in vras-in pi-ne formoj-ne
39
Manchester Phonology Meeting, May 20-22, 2004 Albanian Word Stress
References
G.L. Bevington. Albanian Phonology. Otto Harrassowitz: Wiesbaden, 1974.
Paul de Lacy. The formal expression of markedness. PhD thesis, University of Massachusetts, Amherst, 2002.
A. Dodi and J. Gjinari. Fonetika, volume II of Fonetika dhe Gramatika e gjuhes se sotme letrare Shqipe. Akademia e RPS teShqiperise, 1983.
Michael Kenstowicz. Sonority-driven stress. Ms.,MIT, Rutgers Optimality Archive 33, 1996.
Kelly-Lynne Maynard. An optimality analysis of Albanian word stress. In Xingzhong Li, Luis Lopet, and Tom Stroik, editors, Papersfrom the 1997 Mid-America Linguistics Conference, pages 94–100, 1997.
Leonard Newmark, Phillip Hubbard, and Peter Prifti. Standard Albanian. Stanford University Press: Stanford, 1982.
Sam Rosenthal and Harry van der Hulst. Weight-by-position by position. NLLT, 17:499–540, 1999.
40